Association between apolipoprotein E (APOE) gene and diabetic peripheral neuropathy: A meta-analysis

Some studies have shown the correlation between E4 type of Apolipoprotein E gene and diabetic neuropathy, however, there are contradicting results in recent studies. The aim of this meta-analysis was to investigate the association between APOE4 gene and diabetic neuropathy. Database of Pubmed, Wuhan University Library, and Google were searched. A total of 56 studies were screened and 5 studies were selected for quantitative meta- analysis. This meta-analysis was performed to assess heterogeneity and combine results by using software RevMan 5.2. Strength and relationship were calculated as the odds ratio (OR) with a confidence interval (CL) of 95% using generic inverse variance data. Sensitivity analysis and Publication Bias analysis was conducted. A total of 5 studies were included for quantitative analysis with total of 914 patients (498 diabetic patients with neuropathy as case and 416 diabetic patients without neuropathy as controls). The result of our study indicated that APOE4 carriers were likely to have high risk of diabetic neuropathy. (E4 vs. others; pooled OR with 95% confidence interval 1.80 (1.28, 2.53), p< 0.05 (p =0.0007)). In conclusion, this meta-analysis suggests that there is significant association between APOE4 gene and occurrence of diabetic peripheral neuropathy. However, further studies are needed for more detailed analysis.


Introduction
There is a rising worldwide prevalence of diabetes, which is one of the most challenging health problems in the 21st century. Globally, an estimated 422 million adults were living with diabetes in 2014, compared to 108 million in 1980 with prevalence rising from 4.7% to 8.5% in the adult population (1). American Diabetes Association (ADA) states that by the year 2050, one in three American adults will have diabetes if current trends continue and estimates a person diagnosed at age 50 dies six years earlier than a person without diabetes. According to IDF Atlas, nephropathy, retinopathy and neuropathy are most common microvascular complications of diabetes affecting patient's quality of life and increasing economic burden. However, if early detection of risk factors can be identified then severe consequences like end-stage renal failure, blindness and diabetic foot can be delayed or even prevented (2). It is well known that changes in small arteries and capillaries are not only responsible for micro vascular complications, but also manifests cardiovascular complications in T2DM (3). Diabetic neuropathy is a common complication of diabetes. It has been estimated that approximately 50% of all diabetic patients will develop neuropathy in their lifetime (4). Severe neuropathies may lead to consequences like physical disability, amputations, and cardiac interventions causing burden on public health system and affecting quality of life of patients (5). The most common neuropathies are diabetic peripheral neuropathy (DPN) and cardiovascular autonomic neuropathy (CAN) (6). These neuropathies in patients having diabetes are diverse in character of their symptoms, neurological involvement, risk factors, underlying mechanisms and pathology. Hence clinical presentation and natural history of diabetic peripheral neuropathy are not entirely predictable based on diabetes duration and degree of glycaemic control. In order to target the effective diagnosis, treatment and identification of risk factors for diabetic neuropathies, more researches are needed including study of new genetic variants that can significantly contribute to these conditions. One of the candidate gene linked with these neuropathies is APOE gene. Association of APOE gene with cardiovascular disease and neurological disorders of the central and peripheral nervous system mainly Alzheimer's disease has been widely accepted (7). Studies have also shown positive association of APOE gene with diabetic nephropathy (8). However, significance of APOE gene in diabetic neuropathy is still in debate and inconclusive. APOE was discovered in the early 1970s as a protein component of triglyceride rich lipoproteins (9). APOE is mapped at chromosome 19 and is a polymorphic gene containing three major alleles (E2, E3, and E4) with six possible genotypes (E2/2, E2/3, E2/4, E3/3, E3/4, and E4/4). APOE gene is responsible for producing a protein of 299 amino acids with three different isoforms (e2, e3, e4) differing in two amino acid residues at positions 112 and 158 (10). The APOE3 isoform is the most common isoform with a frequency of approximately 70-80% containing a cysteine at position 112 and an arginine at position 158, while e2(5-10%) possesses cysteines at both positions and e4(10-15%) possesses arginine at both positions and these two isoforms have been thought to be dysfunctional (9,10). Role of APOE polymorphism as risk factor for diabetes neuropathy has been interest of research since many years. However, results from studies vary from each other. The first study was done by Tsuzuki et al in 1998 showing frequency of diabetic neuropathy higher in APOE4 than APOE2 and APOE3 (p < 0.05). However, some studies contradict with the result. Therefore, we performed this meta-analysis to provide a reliable evaluation of the association between APOE4 carriers and diabetic neuropathy.

Data search strategy and sources
We evaluated all case-control studies that studied the association of Apolipoprotein E gene polymorphism with diabetic neuropathy. Data sources were from Pubmed, Wuhan University Library Database and Google. Keywords combinations used were "APOE", "Apolipoprotein E", "Diabetic Neuropathy", "DPN", "NDPN", "Polymorphism", "Genotypes", and "Alleles". Screenings of references of selected articles were also screened. Review articles relevant to our study were also searched in addition to eligible studies. We included all published/unpublished articles written in any language to avoid literature bias. No quality scoring was done.

Study Inclusion and Exclusion Criteria
The studies, which met the following criteria were included for the study. 1) The studies aimed to identify association between APOE polymorphism and diabetic neuropathy (diabetes of any type); 2) Only case-control studies; 3) The studies used diabetic peripheral neuropathy patients as case and diabetic without neuropathy patients as controls; 4) Only human studies. Studies were excluded if: 1) Studies were review articles; 2) Studies conducted on animals. 3) Studies done on other form of Apolipoprotein other than Apolipoprotein E. 4) Unable to contact the original author of unpublished articles.

Data Collection
Two investigators (Dr. Prateek Rajkarnikar and Dr. Arun Bhattarai) independently extracted data after analyzing eligible criteria for relevant studies using standardized data extraction table. The following information were extracted from each study: author name, journal of publication, year of publication, country of study done, APOE gene distribution in each group, APOE gene identification methods and severity assessment methods. Any disagreements were resolved by mutual discussions.

Statistical Analysis
This study compared the APOE gene carriers between diabetic peripheral neuropathy (DPN+) and diabetic without peripheral neuropathy (DPN-) as case and controls respectively by using the odds ratio (OR). We grouped the genotypes and calculated allele carriers as E2 (E2/2+E2/3), E3 (E3/3) and E4 (E3/4+E4/4). We excluded the data of E2/4 because both E2 allele and E4 allele have different effects on occurrence of DPN. It might not be statistically correct to calculate this genotype as E2 carrier or E4 carrier independently. Hardy-Weinberg equilibrium (HWE) test was conducted in the control groups for each study, except for Bedlack et al. due to lack of sufficient data. Chi-square value was calculated and p value >0.05 was considered consistent with HWE. We estimated between-study heterogeneity through all eligible comparisons by using the Chi 2 based Q statistic (11). Heterogeneity was considered significant if p value <0. 10. I 2 test was done to further evaluate heterogeneity with value more than 75%, 50%, and 25% were considered evidence of high, moderate and low statistical heterogeneity respectively (12). Both fixed and random effects model were used. Data were combined using fixed-effects models (when there was no heterogeneity) with inverse-variance weights (13). Random effects are more suitable when there is heterogeneity. The odds ratio (OR) and its 95% confidence interval (CI) was used as the metric of choice. A standard X 2 test was also performed to find out the significance of the overall OR with p value ≤0.05 (14

Results
Data searched on Pubmed, Wuhan University library database and Google initially yielded 56 articles. We performed screening to evaluate the inclusion and exclusion criteria. 11 articles were instantly excluded due to animal reports. Out of 45 articles, 23 articles were excluded because they studied different forms of Apolipoprotein other than APOE polymorphisms. Thirteen articles were excluded because they were not case control studies. Out of nine remaining articles, two were excluded because of inappropriate and incomplete data (15,16). One thesis paper was excluded due to lack of authentication (17) and one study was excluded because it did not have diabetic patients without peripheral neuropathy as controls (18). Finally, five studies were included for this meta-analysis. Four of these studies were published articles (19)(20)(21)(22) whereas, one article was a thesis research paper, which we included for the analysis with the permission of the author (23). The information related to first author, year of publication, country, sample size of case (DPN+) and control (DPN-) for each APOE genotype with allele carriers are presented in table 1.
Two studies were written in English, one study written in Japanese (22), one in Greek (23) and one in Chinese language (20). The identified studies contained total of 914 patients distributed in two groups, diabetic patients with peripheral neuropathy (DPN+) group (n=498) and diabetic patients without peripheral neuropathy (DPN-) group (n=416). The aim in comparison of DPN+ vs. DPN-was to evaluate the role of APOE4 gene on the occurrence of peripheral neuropathy in diabetic patients. The overall effects pooled OR of APOE4 carriers showed statistical significance. 1.80 (1.28, 2.53), p value< 0.05 (p =0.0007). A Forrest plot of comparison is shown in figure  2.
There was no heterogeneity between pooled studies (I 2 <25% P value for heterogeneity > 0.10) and therefore, fixed-effect model was used. We used Begg and Mazumdar test 24 for visual inspection of the funnel plots to see any asymmetry suspecting publication bias. We also used Egger's weighted regression test to statistically assess the publication bias; p value > 0.05 was considered to represent absence of publication bias. The results of both tests showed no evidence of publication bias with Begg's test p value=1 and Egger's test p value = 0.775 shown in Figure  3, 4. One study was deleted each time to see the influence of the individual study on the pooled ORs. However, the corresponding pooled ORs did not change significantly suggesting that our results are statistically correct. This study is the first meta-analysis done to identify association of APOE4 gene with diabetic neuropathy. Although few case control studies had been done in relation to this topic, we believe that our results have statistical significance and may encourage for more studies to be done in relevant to this report.

Discussion
Asymptomatic phase before the actual onset of diabetic hyperglycemia and clinical diagnosis is characteristic of T2DM. The onset of T2DM is usually slow and many years may pass before clinical diagnosis (25). This asymptomatic phase might last for 4 to 7 years in 30-50% patients who remain undiagnosed (26). Diabetic neuropathy is a common complication of diabetes. While estimates vary, depending on the methods and criteria used to diagnose diabetic neuropathy, approximately 50% of all diabetic patients will develop neuropathy in their lifetime (4). While actual etiology of DPN is not clearly understood, gene expression changes, molecular transport, inflammation, oxidative stress due to chronic hyperglycemia, hypertension, dyslipidemia, insulin resistance, and uremia are considered as the key factors contributing to its' pathogenesis (5). However, patient history and clinical presentation based on diabetes duration and degree of glycaemic control cannot entirely predict DPN. It is interesting to note that how some patients develop minimal symptoms of neuropathy after many years of diabetes and others suffer severe nerve damage as early as during pre-diabetes. This inconsistency tends to question the effect of genetic variability in pathogenesis of DPN. Identification of susceptibility genes and their association with diabetes and its complication had been studied since many years. One of the candidate gene linked with these neuropathies is APOE gene. APOE play important role in formation of many plasma-lipoprotein lipid particles like very low-density lipoproteins (VLDL), intermediate density lipoproteins (IDL), high density lipoproteins (HDL) and chylomicrons and helps in the stability and solubility of these lipoproteins during their circulation (10). Association of APOE gene with cardiovascular disease and neurological disorders of the central and peripheral nervous system mainly Alzheimer's disease have been widely accepted (7). Studies have also demonstrated as positive association of APOE gene with diabetic nephropathy (8).
However, significance of APOE gene in diabetic neuropathy is still in debate and inconclusive. Some of the popular studies and their results are described below. This is the first meta-analysis to simply analyze the association of APOE gene with incidence of diabetic peripheral neuropathy. However, there were many limitations. Firstly, there were limited number of studies and results of some included studies were concluded from a relatively small sample size. Secondly, we were unable to examine the interactions among APOE gene and other factors including other candidate genes, diabetes control, environment, which might also play an important role on development of diabetic peripheral neuropathy. Finally, we could not analyze the data to study the APOE gene and its relation with severity of DPN because all authors had used different severity scoring system. We tried to include all possible studies without restriction of language and publication to limit the literature bias. Further studies are needed with detailed comparisons of different APOE genotypes with other clinical variables using common DPN severity scoring system. In conclusion, in spite of several limitations, our metaanalysis suggests that there is significant association between APOE4 gene and occurrence of diabetic peripheral neuropathy. More studies are needed to provide detailed information for evaluating association of APOE gene in DPN.